CT scanners form images based on measurement of X-ray attenuation along multiple paths through the body of a subject. The body is generally irradiated from one side by an X-ray tube, driven by a high-voltage power supply (HVPS), as is known in the art. The X-rays are received by detectors on the opposite side of the body from the tube, which detectors generate signals proportional to the attenuated radiation flux that is incident thereon. The tube revolves around the body (along with the detectors, in third-generation scanners), so that attenuation signal data may be acquired from multiple angular "views." These data are pre-processed, filtered and back-projected to reconstruct an image of a cross-sectional slice is through the body. The body is translated axially relative to the plane of revolution of the tube, so that multiple image slices may be reconstructed, thereby producing a three-dimension CT image.
The quality of the CT image is dependent on the signal/noise ratio (SNR) of the attenuation data, which ratio generally increases with increasing radiation flux at the detectors, particularly when quantum fluctuations are the primary noise source. Therefore, the HVPS is commonly set to apply a relatively high current to the X-ray tube, resulting in a high X-ray flux irradiating the body. High flux is needed especially when imaging areas of the body in which a particularly high SNR is desired, such as the brain, or areas in which the X-ray attenuation is particularly high, such as the pelvis or shoulders, as well as in imaging the bodies of large subjects.
In other parts of the body or along different angular views through the body, however, the X-ray attenuation may not be so high, so that acceptable SNR can be attained at a relatively lower X-ray flux. In these body parts and views, the use of a higher X-ray flux than necessary is undesirable for several reasons: It exposes the subject to excessive radiation dosage, increases wear on the X-ray tube and HVPS, and adds to the cost of operating the CT scanner. Therefore, the X-ray flux should preferably be adjusted to account for the relative attenuation of the body.
Typically, the X-ray attenuation of the body increases with increasing body thickness, i.e., with increasing path length through the body. Because the cross-section of the torso is roughly elliptical, rather than circular, the path length of X-rays traversing the torso from side to side will be substantially greater than the path length from front to back. Therefore, the X-ray tube current that gives an irradiation level appropriate for views in which the tube is near the horizontal, so that the X-rays pass through the torso in the "thick" direction, will be greater than that needed for views in which the tube is near the vertical, in the "thin" direction. Operating the tube at a constant current causes either the thin views to be over-irradiated, or the thick views to be under-irradiated. Moreover, for an asymmetrical body, the noise contribution to the reconstructed image is distributed anisotropically and thus may create image artifacts.
Although the torso is thicker horizontally than vertically, it will be appreciated that other parts of the body, for example, the head, are thicker vertically than horizontally. Furthermore, X-ray attenuation along a given path depends not only on the path length through the body, but also on the types of tissues that the X-rays traverse along the path. Bone in particular attenuates X-rays much more strongly that soft tissue. Any method of compensating for thickness variations must take such differences into account.
It will also be understood that in the context of the present patent application, the terms "thickness," "thick" and "thin" used in reference to directions of radiation flux through the body refer to the combined effects of path length through the body and tissue type in determining directions of greater and lesser attenuation.
U.S. Pat. No. 5,450,462, to Toth et al., which is incorporated herein by reference, describes a CT scanner in which the current applied to the X-ray tube is varied during the revolution of the tube so as to provide greater irradiation in the thick direction than in the thin direction and generally equalize the average flux incident on the detectors in thick and thin directions. Before performing the CT scan, the X-ray path length through the subject's body as a function of view angle is estimated by performing two perpendicular, planar, longitudinal pre-scans of the body. This technique itself, however, increases the time for acquiring a CT image and adds to the radiation exposure of the subject, since ordinarily no more than one such planar pre-scan in performed, if any. It is also inexact, since it takes into account only two planar projections of the body, while the CT scan includes views from a range of angles all around the body.
U.S. Pat. No. 5,485,494, to Williams et al., which is also incorporated herein by reference, describes a technique for varying the X-ray tube current during revolution of the tube by pre-loading the HVPS with a look-up table of X-ray tube current values. This technique is applicable primarily to the pre-scanning method described in the above-mentioned '462 patent and does not overcome the shortcomings pointed out above in this regard.